But who learns what? On the risks of knowledge accumulation through networked learning in R&D

Abstract

In big companies, managerial activities and organizational boundaries will over time hide most unevenly developed skill and knowledge distribution patterns; studying the organizations with the means of modern applied physics is thus quite difficult. People are forced to communicate along the organizational lines, and their personal preferences that could affect the communication networks are often dampened to nearly obsolete. In small companies, however, as well as other less structured non-business organizations, many network patterns exist, based on the preferred cooperation and communication behaviour of human beings, and are observable in various real-life situations. Given their free choice of either to solve the problem themselves or go to one of the colleagues to ask for help, and a preference based on the transactive memory of the organization (a word-of-mouth "reputation" information about who has the skill needed to solve the problem, or who solved the previous one with some similarity) will over time lead to most difficult problems always being solved by one or two key individuals. This paper tests this idea with an agent model to confirm the accumulation of critical knowledge to few individuals. Furthermore, the paper presents a network relation study in a 45-person software solution company. It seems the knowledge is on its way to become distributed according to power law – centralized more and more to a couple of individuals – also in the reality of this case company, even if there are not enough interactions in the five-year history of the company to prove this in a statistically significant way.